CN113761134A - User portrait construction method and device, computer equipment and storage medium - Google Patents
User portrait construction method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN113761134A CN113761134A CN202111089416.1A CN202111089416A CN113761134A CN 113761134 A CN113761134 A CN 113761134A CN 202111089416 A CN202111089416 A CN 202111089416A CN 113761134 A CN113761134 A CN 113761134A
- Authority
- CN
- China
- Prior art keywords
- index value
- index
- user
- evaluation
- evaluation index
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010276 construction Methods 0.000 title claims abstract description 28
- 238000011156 evaluation Methods 0.000 claims abstract description 330
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 230000007774 longterm Effects 0.000 claims description 123
- 230000006399 behavior Effects 0.000 claims description 43
- 238000010606 normalization Methods 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 20
- 238000004140 cleaning Methods 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000012896 Statistical algorithm Methods 0.000 claims description 5
- 238000007405 data analysis Methods 0.000 abstract description 2
- 238000013473 artificial intelligence Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/18—File system types
- G06F16/182—Distributed file systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
Abstract
The invention relates to the technical field of data analysis, in particular to a user portrait construction method and device, computer equipment and a storage medium. The method comprises the following steps: setting at least two judgment indexes related to the target type; acquiring a judgment index value of a target user on the judgment index through a big data platform; calculating the judgment index value through a TF-IDF algorithm to obtain the judgment index weight; normalizing the evaluation index value to obtain a normalized evaluation index value; calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index; and constructing a user portrait of the target user according to the index value of the target type. The method calculates the index value of the target type through the normalized judgment index value and the judgment index weight of the judgment index, constructs the user portrait according to the index value, and can improve the accuracy of the user portrait.
Description
Technical Field
The invention relates to the technical field of data analysis, in particular to a user portrait construction method and device, computer equipment and a storage medium.
Background
With the development of artificial intelligence, user image models are widely applied in various fields. For the establishment of the user portrait model, generally, a plurality of indexes are combined for comprehensive evaluation, and after the indexes for evaluation are selected, how to set the weights of the indexes becomes a problem to be solved.
At present, the weight design aiming at indexes in the industry is mainly divided into a subjective weighting method and an objective weighting method, wherein the subjective weighting method is mainly judged subjectively according to the experience of experts or decision makers, the method is mature, but has poor objectivity and is mainly related to the experience of the decision makers, the subjective opinion is large, and a plurality of objective factors cannot be considered comprehensively; the objective weighting method is mainly obtained according to data statistics, is high in objectivity, avoids the influence of artificial subjective factors, but due to lack of comprehensiveness, the situation that the calculated index weight is inconsistent with the importance of the index possibly occurs, and the accuracy of the constructed user portrait is low.
Disclosure of Invention
In view of the above, it is desirable to provide a user portrait construction method, device, computer device and storage medium to solve the problem that the existing index weight is difficult to satisfy both the objectivity and the comprehensiveness of the index, and to improve the precision of the user portrait.
A user representation construction method, comprising:
receiving an index setting instruction, and setting a target type of a user portrait and at least two judgment indexes related to the target type according to the index setting instruction;
acquiring a judgment index value of a target user on the judgment index through a big data platform;
calculating the judgment index value through a TF-IDF algorithm to obtain a judgment index weight corresponding to the judgment index;
carrying out normalization processing on the evaluation index value to obtain a normalized evaluation index value;
calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index;
and constructing the user portrait of the target user according to the index value of the target type.
A user representation construction apparatus comprising:
the judgment index setting module is used for receiving an index setting instruction and setting at least two judgment indexes related to the target type according to the index setting instruction;
the evaluation index value module is used for acquiring the evaluation index value of the target user on the evaluation index through a big data platform;
the evaluation index weight module is used for calculating the evaluation index value through a TF-IDF algorithm to obtain the evaluation index weight corresponding to the evaluation index; wherein, one judgment index corresponds to one judgment index value;
the normalization evaluation index value module is used for performing normalization processing on the evaluation index value to obtain a normalization evaluation index value;
the index value module is used for calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index;
and the user portrait module is used for constructing the user portrait of the target user according to the index value of the target type.
A computer device comprising a memory, a processor and computer readable instructions stored in the memory and executable on the processor, the processor implementing the user representation construction method when executing the computer readable instructions.
One or more readable storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform a user representation construction method as described above.
According to the user portrait construction method, the user portrait construction device, the computer equipment and the storage medium, the index setting instruction is received, and at least two judgment indexes related to the target type are set according to the index setting instruction; acquiring a judgment index value of a target user on the judgment index through a big data platform; calculating the judgment index value through a TF-IDF algorithm to obtain a judgment index weight corresponding to the judgment index; wherein, one judgment index corresponds to one judgment index value; carrying out normalization processing on the evaluation index value to obtain a normalized evaluation index value; calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index; and constructing the user portrait of the target user according to the index value of the target type. The invention can solve the problem that the existing index weight is difficult to set and the objectivity and the comprehensiveness of the index are difficult to be considered, and improves the accuracy of user portrayal.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments of the present invention will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without inventive labor.
FIG. 1 is a diagram illustrating an application environment of a user representation construction method according to an embodiment of the present invention;
FIG. 2 is a flow chart illustrating a user representation construction method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a user representation creation apparatus according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a computer device according to an embodiment of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result. The embodiment can acquire the index data of the target user through the special artificial intelligence chip.
The user portrait construction method provided by this embodiment can be applied to the application environment shown in fig. 1, in which the client communicates with the server. The client includes, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server can be implemented by an independent server or a server cluster composed of a plurality of servers. The server may be an independent server, or may be a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a Content Delivery Network (CDN), and a big data and artificial intelligence platform.
In an embodiment, as shown in fig. 2, a user portrait construction method is provided, which is described by taking the application of the method to the server side in fig. 1 as an example, and includes the following steps:
s10, receiving an index setting instruction, and setting a target type of the user portrait and at least two judgment indexes related to the target type according to the index setting instruction.
Understandably, the index setting instruction may be an instruction formed by user input. The target type can be set according to actual needs, such as financing, game, social contact and shopping. Each target type is associated with at least two judgment indicators. For example, when the target type is financing, the associated evaluation index may be personal assets, personal academic calendar, financing product purchase data, financing product page access data, and the like.
The evaluation index is obtained by decomposing the target type based on an analytic hierarchy process. Specifically, a hierarchical structure is constructed through an analytic hierarchy process, and the target type is decomposed into a user long-term attribute and a user short-term attribute according to a time span. The user long-term attributes comprise user self attributes and user long-term behavior attributes. The user short-term attributes include user short-term behavior attributes. The user's own attributes include inherent attributes such as the user's gender, age, academic calendar, and the like. The user long-term behavior attributes include long-term interests of the user, for example, a user's favorite drama in the last year. The user short-term behavior attributes comprise recent behavior activities of the user, for example, a user likes to manage money in a month. And determining an index for judgment according to the long-term attribute and the short-term attribute of the user. After determining the index for evaluation, at least two evaluation indexes associated with the target type are set according to the index setting instruction. For example, the evaluation index may be the number of clicks of a user on a financial product, and the index value of the evaluation index, which is the number of clicks, may be calculated as n times per month. The evaluation indexes comprise long-term evaluation indexes and short-term evaluation indexes. The long-term attributes of the user correspond to the long-term evaluation indicators. The user short-term attributes correspond to short-term evaluation indicators.
And S20, acquiring the judgment index value of the target user on the judgment index through a big data platform.
It is understood that the target user refers to a user corresponding to the target type. For example, the target user is a user who likes financing. And the big data platform is a data processing platform for storing a large amount of user data. The big data platform comprises a spark (computing engine) framework, spark SQL and hadoop (distributed computing) platforms, and can be used for performing designated analysis processing on user data to obtain corresponding data. Wherein sparkSQL is a spark component used for processing structured data. The spark framework is mature and stable, mass data can be calculated quickly, and complex indexes can be calculated quickly.
When obtaining the evaluation index value corresponding to the evaluation index, the evaluation index value may be collected and stored to an hdfs (Hadoop distributed file system) storage medium of a Hadoop (distributed computing) platform by using an sql technique, and the hdfs storage medium may include user behavior attribute data from an oracle database and buried point data stored in the hdfs storage medium. The sqoop is a source-opening tool, and is mainly used for data transmission between a Hadoop (hive) and a traditional database, and data in a relational database can be imported into hdfs of the Hadoop and can also be imported into the relational database.
And S30, calculating the judgment index value through a TF-IDF algorithm to obtain the judgment index weight corresponding to the judgment index.
Understandably, one judgment index corresponds to one judgment index value. The TF-IDF algorithm is a weighting technique for information retrieval and data mining, and is generally used to evaluate the importance of a word to one of a set of files or a corpus of files. The TF-IDF algorithm includes:
wherein, TiAll judgment indexes of a certain target user;
Pjall target users;
w (P, T) represents a criterion T used to mark the data value of the user P;
TF (P, T) represents the proportion of W (P, T) in all judgment indexes of the user P;
∑∑W(Pj，Ti) Data values representing all judgment indicators of all users;
∑W(Pjt) represents the sum of all users of the evaluation index T;
IDF (P, T) represents the occurrence probability of the evaluation index T among all the evaluation indexes;
TF (P, T) × IDF (P, F) is the weighted value of the target user P on the evaluation index T.
In one example, Pj∈(U1、U2、U3)，TiE (A, B, C, D), user U as shown in Table 11There are 5 judgment indexes A and B2, 1 judgment index C. From TF (P, T), user U can be computed1The TF value on the criterion index a is TF 5/(5+2+ 1). Further, from IDF (P, T), user U is calculated1IDF value on evaluation index A. After obtaining the TF value and the IDF value, calculating the user U according to TF (P, T) IDF (P, F)1Weight value on the evaluation index A.
TABLE 1 user index statistics
| A | B | C | D | |
| U1 | 5 | 2 | 1 | 0 |
| U2 | 4 | 10 | 7 | 6 |
| U3 | 6 | 1 | 8 | 2 |
Specifically, the evaluation index value of the target user on each evaluation index can be obtained through the big data platform. One judgment index corresponds to one judgment index value. After obtaining the evaluation index value, calculating the evaluation index weight of each evaluation index through a TF-IDF algorithm according to the evaluation index value on each evaluation index.
S40, carrying out normalization processing on the judgment index value to obtain a normalized judgment index value.
Understandably, the evaluation index value can be acquired through a big data platform. Normalization refers to scaling the data to fall within a small specific interval. Since the measurement units of the indexes are different, in order to be able to participate in the calculation of the indexes, the indexes need to be normalized, and the values of the indexes are mapped to a certain value interval through function transformation.
Specifically, the judgment index value is normalized through a preset normalization model to obtain a normalized judgment index value, and the judgment index value is made to fall into a (0, 1) interval. Wherein the preset normalization model may be a min-max model (min-max normalization model) or a z-score model (normal distribution normalization model). The normalization can make the indexes of different dimensions in the same numerical order, reduce the influence of the judgment index with larger variance on the user portrait, make the user portrait more accurate and accelerate the convergence speed of the learning algorithm.
S50, calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index.
Understandably, one judgment index corresponds to one normalized judgment index value, and one judgment index corresponds to one judgment index weight. The evaluation index includes a long-term evaluation index and a short-term evaluation index. The long-term attributes of the user correspond to the long-term evaluation indicators. The user short-term attributes correspond to short-term evaluation indicators.
Specifically, according to the short-term evaluation index, the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index are obtained for weighted calculation, and a first index value corresponding to the short-term attribute of the user is obtained. And acquiring the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index according to the long-term evaluation index, and performing weighted calculation to obtain a second index value corresponding to the long-term attribute of the user. And adding the first index value and the second index value to obtain the index value of the target type.
And S60, constructing the user portrait of the target user according to the index value of the target type.
Understandably, the long-term attribute and the short-term attribute of the user are weighted and calculated to finally obtain the index value of the target type, and the index value is stored in a hive data table of a hadoop (distributed computing) platform for personalized recommendation. Wherein hive is a data warehouse tool based on Hadoop.
The target type index value is the final result of the user representation, for example, the interest score of the a user for the gun shot game is 0.9, and the interest score of the narrative novel is 0.6. The role of the evaluation index is only one means of calculating the value of interest.
Receiving an index setting instruction, and setting at least two judgment indexes associated with the target type according to the index setting instruction in steps S10-S60; acquiring a judgment index value of a target user on the judgment index through a big data platform; calculating the judgment index value through a TF-IDF algorithm to obtain a judgment index weight corresponding to the judgment index; wherein, one judgment index corresponds to one judgment index value; carrying out normalization processing on the evaluation index value to obtain a normalized evaluation index value; calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index; and constructing the user portrait of the target user according to the index value of the target type. The method calculates the index value of the target type through the normalized judgment index value and the judgment index weight of the judgment index, constructs the user portrait according to the index value, and can improve the accuracy of the user portrait.
Optionally, the evaluation index includes a short-term evaluation index and a long-term evaluation index; and the target type is divided into a user short-term attribute and a user long-term attribute according to the time span. In step S50, the calculating the index value of the target type according to the normalized evaluation index value and the evaluation index weight of each evaluation index includes:
s501, acquiring the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index;
s502, calculating a first index value corresponding to the user short-term attribute according to the normalized judgment index value corresponding to the short-term judgment index and the judgment index weight;
s503, acquiring the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index, and calculating a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index;
s504, adding the first index value and the second index value to obtain the index value of the target type.
Understandably, the target types are divided into user short-term attributes and user long-term attributes in time span. The short-term evaluation index corresponds to the user short-term attribute. For example, if a user likes to manage money in the last month, the short-term evaluation index may be the number of times the user clicks on a money management product or the number of visits. One short-term evaluation index corresponds to one normalized evaluation index value. One short-term evaluation index corresponds to one evaluation index weight. According to the normalized evaluation index values and the evaluation index weights corresponding to the short-term evaluation indexes, a user short-term attribute value of the user short-term attribute, namely a first index value, can be calculated. The long-term evaluation index corresponds to a long-term attribute of the user. For example, if a user likes to manage money in the last year, the long-term evaluation index may be the number of times the user clicks on a money management product or the number of visits. One long-term evaluation index corresponds to one normalized evaluation index value. One long-term evaluation index corresponds to one evaluation index weight. And calculating a user long-term attribute value of the user long-term attribute, namely a second index value according to the normalized evaluation index values and the evaluation index weights corresponding to the plurality of long-term evaluation indexes. And adding the first index value and the second index value to obtain the index value of the target type. The user long-term attributes comprise self attributes and user long-term behavior attributes. The user long-term attribute value comprises a user self attribute value and a user long-term behavior attribute value.
In steps S501-S504, by giving different weights to the evaluation indexes of the users at different periods, the behaviors of the users at different periods are perfectly distinguished, and the accuracy of the user portrait is improved.
Optionally, the user long-term attribute includes a user long-term behavior attribute and a user self attribute. In step S503, namely, the normalization evaluation index value and the evaluation index weight corresponding to the long-term evaluation index are obtained, and the normalization evaluation index value and the evaluation index weight corresponding to the long-term evaluation index are weighted to obtain a second index value corresponding to the long-term attribute of the user; the user long-term attributes comprise user long-term behavior attributes and user self attributes, and comprise:
s5031, acquiring a self-evaluation index value and a self-evaluation index weight corresponding to the user self-attribute, and a long-term behavior evaluation index value and a long-term behavior evaluation index weight corresponding to the user long-term behavior attribute;
s5032, performing weighted calculation on the self-evaluation index value and the self-evaluation index weight, and the long-term behavior evaluation index value and the long-term behavior evaluation index weight to obtain a second index value corresponding to the long-term attribute of the user.
Understandably, the user long-term attributes include self attributes and user long-term behavior attributes. The normalized evaluation index value corresponding to the long-term evaluation index comprises a self evaluation index value corresponding to the self attribute of the user and a long-term behavior evaluation index value corresponding to the long-term behavior attribute of the user. The user's own attributes refer to the inherent attributes of the user, such as the user's age, gender, income, academic history, and the like. The self-evaluation index value refers to an evaluation index value corresponding to the user's own attribute. The user long-term behavior attribute comprises long-term behavior activities of the user, such as the number of clicks or purchases of financial products by the user in the last year. The long-term behavior evaluation index value refers to an evaluation index value corresponding to the long-term behavior attribute of the user. For example, the long-term evaluation index is the number of monthly clicks on the financial product, and if the number of monthly clicks on the financial product within 3 months of the user is 30, the number of monthly clicks on the financial product by the user is 10 times/month, and 10 times/month is the long-term behavior evaluation index value.
Specifically, after obtaining the self-evaluation index value of the user self-attribute, the evaluation index weight corresponding to the user self-attribute, the long-term behavior evaluation index value of the user long-term behavior attribute, and the evaluation index weight corresponding to the user long-term behavior attribute, the self-evaluation index values of the user self-attributes are weighted and calculated according to the evaluation index weight corresponding to the user self-attribute, and the user long-term behavior attribute value is obtained. And performing weighted calculation on the long-term behavior evaluation index values of the long-term behavior attributes of the plurality of users according to the evaluation index weight corresponding to the long-term behavior attributes of the users to obtain the self attribute values of the users. And then, summing the attribute value of the user and the long-term behavior attribute value of the user to obtain a second index value corresponding to the long-term attribute of the user.
Optionally, in step S20, the obtaining of the evaluation index value of the evaluation index through the big data platform includes:
s201, acquiring index initial data of the judgment index through a big data platform;
s202, cleaning the initial index data by a preset cleaning method to obtain index cleaning data;
s203, processing the index cleaning data through a preset statistical algorithm to obtain a judgment index value of the judgment index.
Understandably, the index initial data is data directly acquired on a big data platform according to the judgment index. Since these data may include garbage data such as abnormal data and test data, which may affect the analysis processing result, these garbage data need to be cleaned and eliminated. Where the test data includes data generated by the tester for testing the system during the time the system is on-line, such as clicks that like classical literature, but are not really liked because such things are never seen at all. The abnormal data includes data generated by abnormal operation of a system or a user, for example, the user clicks a certain character bar tens of thousands of times. The preset cleaning method includes, but is not limited to, python language technology. The pre-set statistical algorithm may be an algorithm using a spark (computing engine) based framework, spark sql, and hadoop platform. Wherein sparkSQL is a spark component used for processing structured data.
Specifically, after index initial data of the judgment index is obtained through the big data platform, the index initial data is cleaned according to a preset cleaning method, the garbage data is removed, and index cleaning data is obtained after cleaning. And then, processing the index cleaning data through a preset statistical algorithm to obtain a judgment index value corresponding to the judgment index. One judgment index corresponds to one judgment index value.
In steps S201 to S203, the accuracy of the evaluation of the index value can be improved by cleaning the initial index data, so as to improve the accuracy of the user image.
Optionally, the evaluation index includes a short-term evaluation index and a long-term evaluation index; and the target type is divided into a user short-term attribute and a user long-term attribute according to the time span. In step S50, the calculating the index value of the target type according to the normalized evaluation index value and the evaluation index weight of each evaluation index further includes:
s505, acquiring the normalized judgment index value and the judgment index weight corresponding to the short-term judgment index;
s506, calculating a first index value corresponding to the user short-term attribute according to the normalized judgment index value corresponding to the short-term judgment index and the judgment index weight;
s507, acquiring the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index; calculating a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value corresponding to the long-term evaluation index and the evaluation index weight;
s508, acquiring a first expert evaluation weight of the user short-term attribute and a second expert evaluation weight of the user long-term attribute from an expert knowledge domain library;
s509, calculating the index value of the target type according to the first index value, the second index value, the first expert evaluation weight and the second expert evaluation weight.
Understandably, the expert knowledge domain library is used for storing the judgment weight of the long-term attribute of the user and the judgment weight of the short-term attribute of the user in each domain determined by the domain expert according to professional knowledge. The first expert evaluation weight is the evaluation weight of the short-term attribute of the user, and the second expert evaluation weight is the evaluation weight of the long-term attribute of the user. The first index value is a user long-term attribute value, and the second index value is a user long-term attribute value. According to the normalized evaluation index values and the evaluation index weights corresponding to the short-term evaluation indexes, a user short-term attribute value of the user short-term attribute, namely a first index value, can be calculated. And calculating a user long-term attribute value of the user long-term attribute, namely a second index value according to the normalized evaluation index values and the evaluation index weights corresponding to the plurality of long-term evaluation indexes.
Specifically, after a first expert evaluation weight of the short-term attribute of the user and a second expert evaluation weight of the long-term attribute of the user are obtained from the expert knowledge domain library, the first index value and the second index value are weighted and summed according to the first expert evaluation weight and the second expert evaluation weight, and the index value of the target type is obtained.
In steps S505-S509, acquiring a first expert evaluation weight of the user short-term attribute and a second expert evaluation weight of the user long-term attribute from an expert knowledge domain library; and calculating the index value of the target type according to the first index value, the second index value, the first expert evaluation weight and the second expert evaluation weight, wherein the final index value is fused with an expert knowledge field base, so that the accuracy of the index value is improved, and the accuracy of the user portrait is further improved.
Optionally, in step S30, that is, the calculating the evaluation index value by the TF-IDF algorithm to obtain the evaluation index weight corresponding to the evaluation index includes:
s301, determining a first proportion of any one judgment index in all judgment indexes of the target user according to the judgment index value and the judgment index of any one target user;
s302, determining a second occupation proportion of any one judgment index in all judgment indexes of all target users according to the judgment index values and the judgment indexes of all the target users;
s303, determining a judgment index weight corresponding to the judgment index according to the first occupation weight and the second occupation weight.
Understandably, the TF-IDF algorithm includes:
wherein, TiAll judgment indexes of a certain target user;
Pjall target users;
w (P, T) represents a criterion T used to mark the data value of the user P;
TF (P, T) represents the proportion of W (P, T) in all judgment indexes of the user P;
∑∑W(Pj，Ti) Data values representing all judgment indicators of all users;
∑W(Pjt) represents the sum of all users of the evaluation index T;
IDF (P, T) represents the occurrence probability of the evaluation index T among all the evaluation indexes;
TF (P, T) × IDF (P, F) is the weighted value of the target user P on the evaluation index T.
Specifically, according to the evaluation index value and the evaluation index of any one target user, the frequency of any one evaluation index appearing in all the evaluation indexes on the target user, that is, the first occupation ratio, can be determined by the calculation formula TF (P, T). Meanwhile, according to the judgment index values and the judgment indexes of all target users, the frequency of any one judgment index appearing in all the judgment indexes of all the target users can be determined through the calculation formula IDF (P, T), namely the second occupation ratio. After the first and second occupancy ratios are obtained, the evaluation index weight corresponding to the evaluation index can be determined according to the calculation formula TF (P, T) × IDF (P, F).
In one example, Pj∈(U1、U2、U3)，TiE (A, B, C, D), user U as shown in Table 11There are 5 judgment indexes A, 2 judgment indexes B and 1 judgment index C. From TF (P, T), user U can be computed1The TF value on the criterion index a is TF 5/(5+2+ 1). Further, from IDF (P, T), user U is calculated1IDF value on evaluation index A. After obtaining the TF value and the IDF value, calculating the user U according to TF (P, T) IDF (P, F)1Weight value on the evaluation index A.
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
In one embodiment, a user representation construction apparatus is provided, which corresponds to the user representation construction method in the above embodiments one to one. As shown in FIG. 3, the user profile construction apparatus comprises a judgment index setting module 10, a judgment index weight module 20, a normalized judgment index value module 30, an index value module 40 and a user profile module 50. The functional modules are explained in detail as follows:
the judgment index setting module 10 is used for receiving an index setting instruction and setting at least two judgment indexes related to the target type according to the index setting instruction;
the evaluation index value module 20 is configured to obtain an evaluation index value of the target user on the evaluation index through a big data platform;
the evaluation index weight module 30 is configured to calculate the evaluation index value through a TF-IDF algorithm to obtain an evaluation index weight corresponding to the evaluation index; wherein, one judgment index corresponds to one judgment index value;
a normalization evaluation index value module 40, configured to perform normalization processing on the evaluation index value to obtain a normalization evaluation index value;
an index value module 50, configured to calculate an index value of the target type according to the normalized evaluation index value and the evaluation index weight of each evaluation index;
and a user representation module 60 for constructing a user representation of the target user according to the index value of the target type.
Optionally, the metric module 50 includes:
a short-term evaluation index data acquisition unit configured to acquire the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index;
a first index value unit, configured to calculate a first index value corresponding to the user short-term attribute according to the normalized evaluation index value corresponding to the short-term evaluation index and the evaluation index weight;
a second index value unit configured to obtain the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index, and calculate a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index;
and the index value unit is used for summing the first index value and the second index value to obtain the index value of the target type.
Optionally, the second index value unit includes:
a long-term evaluation index data acquisition unit for acquiring a self evaluation index value and a self evaluation index weight corresponding to the user self attribute, and a long-term behavior evaluation index value and a long-term behavior evaluation index weight corresponding to the user long-term behavior attribute;
and the second index value calculation unit is used for performing weighted calculation on the self evaluation index value and the self evaluation index weight, and the long-term behavior evaluation index value and the long-term behavior evaluation index weight to obtain a second index value corresponding to the long-term attribute of the user.
Optionally, the index weight evaluating module 20 includes:
the index initial data unit is used for acquiring the index initial data of the judgment index through a big data platform;
the data cleaning unit is used for cleaning the initial index data through a preset cleaning method to obtain index cleaning data;
and the index value judging unit is used for processing the index cleaning data through a preset statistical algorithm to obtain the judging index value of the judging index.
Optionally, the metric value module 50 further includes:
a short-term evaluation index data acquisition unit configured to acquire the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index;
a first index value unit, configured to calculate a first index value corresponding to the user short-term attribute according to the normalized evaluation index value corresponding to the short-term evaluation index and the evaluation index weight;
a second index value unit configured to obtain the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index, and calculate a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index;
the expert evaluation weight unit is used for acquiring a first expert evaluation weight of the user short-term attribute and a second expert evaluation weight of the user long-term attribute from an expert knowledge domain library;
and the index value unit is also used for calculating the index value of the target type according to the first index value, the second index value, the first expert evaluation weight and the second expert evaluation weight.
Optionally, the evaluation index weight module 20 includes:
a first occupation proportion unit, configured to determine, according to the evaluation index value and the evaluation index of any one of the target users, a first occupation proportion of any one of the evaluation indexes in all the evaluation indexes of the target user;
a second occupancy proportion unit, configured to determine a second occupancy proportion of any one of the evaluation indexes in all the evaluation indexes on all the target users according to the evaluation index values and the evaluation indexes of all the target users;
and the judgment index weight unit is used for determining the judgment index weight corresponding to the judgment index according to the first occupation weight and the second occupation weight.
For the specific limitation of the user representation constructing apparatus, reference may be made to the above limitation of the user representation constructing method, which is not described herein again. The modules in the user representation constructing device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 4. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a readable storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer readable instructions. The internal memory provides an environment for the operating system and execution of computer-readable instructions in the readable storage medium. The network interface of the computer device is used for communicating with an external server through a network connection. The computer readable instructions, when executed by a processor, implement a user representation construction method. The readable storage media provided by the present embodiment include nonvolatile readable storage media and volatile readable storage media.
In one embodiment, a computer device is provided, comprising a memory, a processor, and computer readable instructions stored on the memory and executable on the processor, the processor when executing the computer readable instructions implementing the steps of:
receiving an index setting instruction, and setting a target type of a user portrait and at least two judgment indexes related to the target type according to the index setting instruction;
acquiring a judgment index value of a target user on the judgment index through a big data platform;
calculating the judgment index value through a TF-IDF algorithm to obtain a judgment index weight corresponding to the judgment index;
carrying out normalization processing on the evaluation index value to obtain a normalized evaluation index value;
calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index;
and constructing the user portrait of the target user according to the index value of the target type.
In one embodiment, one or more computer-readable storage media storing computer-readable instructions are provided, the readable storage media provided by the embodiments including non-volatile readable storage media and volatile readable storage media. The readable storage medium has stored thereon computer readable instructions which, when executed by one or more processors, perform the steps of:
receiving an index setting instruction, and setting a target type of a user portrait and at least two judgment indexes related to the target type according to the index setting instruction;
acquiring a judgment index value of a target user on the judgment index through a big data platform;
calculating the judgment index value through a TF-IDF algorithm to obtain a judgment index weight corresponding to the judgment index;
carrying out normalization processing on the evaluation index value to obtain a normalized evaluation index value;
calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index;
and constructing the user portrait of the target user according to the index value of the target type.
It will be understood by those of ordinary skill in the art that all or part of the processes of the methods of the above embodiments may be implemented by hardware related to computer readable instructions, which may be stored in a non-volatile readable storage medium or a volatile readable storage medium, and when executed, the computer readable instructions may include processes of the above embodiments of the methods. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. A user portrait construction method, comprising:
receiving an index setting instruction, and setting a target type of a user portrait and at least two judgment indexes related to the target type according to the index setting instruction;
acquiring a judgment index value of a target user on the judgment index through a big data platform;
calculating the judgment index value through a TF-IDF algorithm to obtain a judgment index weight corresponding to the judgment index;
carrying out normalization processing on the evaluation index value to obtain a normalized evaluation index value;
calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index;
and constructing the user portrait of the target user according to the index value of the target type.
2. The user representation construction method of claim 1, wherein the evaluation indicators comprise short-term evaluation indicators and long-term evaluation indicators; the target type is divided into a user short-term attribute and a user long-term attribute according to the time span;
the calculating the index value of the target type according to the normalized evaluation index value and the evaluation index weight of each evaluation index comprises the following steps:
acquiring the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index;
calculating a first index value corresponding to the user short-term attribute according to the normalized judgment index value corresponding to the short-term judgment index and the judgment index weight;
acquiring the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index; calculating a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value corresponding to the long-term evaluation index and the evaluation index weight;
and adding the first index value and the second index value to obtain the index value of the target type.
3. A user representation construction method according to claim 2, wherein said user long-term attributes include a user long-term behaviour attribute and a user self attribute; the obtaining the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index, and performing weighting processing on the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index to obtain a second index value corresponding to the long-term attribute of the user includes:
acquiring a self-evaluation index value and a self-evaluation index weight corresponding to the user self-attribute, and a long-term behavior evaluation index value and a long-term behavior evaluation index weight corresponding to the user long-term behavior attribute;
and carrying out weighted calculation on the self evaluation index value and the self evaluation index weight as well as the long-term behavior evaluation index value and the long-term behavior evaluation index weight to obtain a second index value corresponding to the long-term attribute of the user.
4. The user representation construction method of claim 1, wherein the obtaining of the evaluation index value of the evaluation index through a big data platform comprises:
acquiring index initial data of the judgment index through a big data platform;
cleaning the initial index data by a preset cleaning method to obtain index cleaning data;
and processing the index cleaning data through a preset statistical algorithm to obtain the judgment index value of the judgment index.
5. The user representation construction method of claim 1, wherein the evaluation indicators comprise short-term evaluation indicators and long-term evaluation indicators; the target type is divided into a user short-term attribute and a user long-term attribute according to the time span;
the calculating the index value of the target type according to the normalized evaluation index value and the evaluation index weight of each evaluation index further comprises:
acquiring the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index;
calculating a first index value corresponding to the user short-term attribute according to the normalized judgment index value corresponding to the short-term judgment index and the judgment index weight;
acquiring the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index; calculating a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value corresponding to the long-term evaluation index and the evaluation index weight;
acquiring a first expert evaluation weight of the short-term attribute of the user and a second expert evaluation weight of the long-term attribute of the user from an expert knowledge domain library;
and calculating the index value of the target type according to the first index value, the second index value, the first expert evaluation weight and the second expert evaluation weight.
6. The method for constructing a user portrait according to claim 1, wherein the calculating the evaluation index value by the TF-IDF algorithm to obtain the evaluation index weight corresponding to the evaluation index comprises:
determining a first proportion of any one judgment index in all judgment indexes of the target user according to the judgment index value and the judgment index of any one target user;
determining a second occupation proportion of any one judgment index in all judgment indexes of all target users according to the judgment index values and the judgment indexes of all the target users;
and determining the judgment index weight corresponding to the judgment index according to the first occupation weight and the second occupation weight.
7. A user representation construction apparatus, comprising:
the judgment index setting module is used for receiving an index setting instruction and setting at least two judgment indexes related to the target type according to the index setting instruction;
the evaluation index value module is used for acquiring the evaluation index value of the target user on the evaluation index through a big data platform;
the evaluation index weight module is used for calculating the evaluation index value through a TF-IDF algorithm to obtain the evaluation index weight corresponding to the evaluation index; wherein, one judgment index corresponds to one judgment index value;
the normalization evaluation index value module is used for performing normalization processing on the evaluation index value to obtain a normalization evaluation index value;
the index value module is used for calculating the index value of the target type according to the normalized judgment index value and the judgment index weight of each judgment index;
and the user portrait module is used for constructing the user portrait of the target user according to the index value of the target type.
8. The user representation construction apparatus of claim 7 wherein the evaluation metrics comprise a short term evaluation metric and a long term evaluation metric; the target type is divided into a user short-term attribute and a user long-term attribute according to the time span; the calculating the index value of the target type according to the normalized evaluation index value and the evaluation index weight of each evaluation index comprises the following steps:
a short-term evaluation index data acquisition unit configured to acquire the normalized evaluation index value and the evaluation index weight corresponding to the short-term evaluation index;
a first index value unit, configured to calculate a first index value corresponding to the user short-term attribute according to the normalized evaluation index value corresponding to the short-term evaluation index and the evaluation index weight;
a second index value unit configured to obtain the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index, and calculate a second index value corresponding to the long-term attribute of the user according to the normalized evaluation index value and the evaluation index weight corresponding to the long-term evaluation index;
and the index value unit is used for summing the first index value and the second index value to obtain the index value of the target type.
9. A computer device comprising a memory, a processor and computer readable instructions stored in the memory and executable on the processor, wherein the processor when executing the computer readable instructions implements a user representation construction method as claimed in any one of claims 1 to 6.
10. One or more readable storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the user representation construction method of any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111089416.1A CN113761134A (en) | 2021-09-16 | 2021-09-16 | User portrait construction method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111089416.1A CN113761134A (en) | 2021-09-16 | 2021-09-16 | User portrait construction method and device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113761134A true CN113761134A (en) | 2021-12-07 |
Family
ID=78796118
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111089416.1A Pending CN113761134A (en) | 2021-09-16 | 2021-09-16 | User portrait construction method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113761134A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190103102A1 (en) * | 2017-10-04 | 2019-04-04 | The Toronto-Dominion Bank | Persona-based conversational interface personalization using social network preferences |
| CN109784671A (en) * | 2018-12-24 | 2019-05-21 | 远光软件股份有限公司 | A kind of user experience quality appraisal procedure and system based on user behavior analysis |
| CN111444428A (en) * | 2020-03-27 | 2020-07-24 | 腾讯科技（深圳）有限公司 | Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium |
| CN111723256A (en) * | 2020-06-03 | 2020-09-29 | 开普云信息科技股份有限公司 | Government affair user portrait construction method and system based on information resource library |
| CN112035541A (en) * | 2020-08-28 | 2020-12-04 | 平安科技（深圳）有限公司 | Client image drawing method and device, computer readable storage medium and terminal equipment |
| CN112308173A (en) * | 2020-12-28 | 2021-02-02 | 平安科技（深圳）有限公司 | Multi-target object evaluation method based on multi-evaluation factor fusion and related equipment thereof |
-
2021
- 2021-09-16 CN CN202111089416.1A patent/CN113761134A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190103102A1 (en) * | 2017-10-04 | 2019-04-04 | The Toronto-Dominion Bank | Persona-based conversational interface personalization using social network preferences |
| CN109784671A (en) * | 2018-12-24 | 2019-05-21 | 远光软件股份有限公司 | A kind of user experience quality appraisal procedure and system based on user behavior analysis |
| CN111444428A (en) * | 2020-03-27 | 2020-07-24 | 腾讯科技（深圳）有限公司 | Information recommendation method and device based on artificial intelligence, electronic equipment and storage medium |
| CN111723256A (en) * | 2020-06-03 | 2020-09-29 | 开普云信息科技股份有限公司 | Government affair user portrait construction method and system based on information resource library |
| CN112035541A (en) * | 2020-08-28 | 2020-12-04 | 平安科技（深圳）有限公司 | Client image drawing method and device, computer readable storage medium and terminal equipment |
| CN112308173A (en) * | 2020-12-28 | 2021-02-02 | 平安科技（深圳）有限公司 | Multi-target object evaluation method based on multi-evaluation factor fusion and related equipment thereof |
Non-Patent Citations (1)
| Title |
|---|
| 王洪伟;邹莉: "考虑长期与短期兴趣因素的用户偏好建模", 《同济大学学报(自然科学版)》, no. 6, 30 June 2013 (2013-06-30), pages 953 - 960 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3819821B1 (en) | User feature generating method, device, and apparatus, and computer-readable storage medium | |
| US20210232951A1 (en) | Systems and methods of processing personality information | |
| US20180247362A1 (en) | Optimized recommendation engine | |
| EP2960849A1 (en) | Method and system for recommending an item to a user | |
| CN108021673A (en) | A kind of user interest model generation method, position recommend method and computing device | |
| CN112231533A (en) | Data processing method, device, equipment and storage medium | |
| CN111967914A (en) | User portrait based recommendation method and device, computer equipment and storage medium | |
| CN111695084A (en) | Model generation method, credit score generation method, device, equipment and storage medium | |
| Saleem et al. | Personalized decision-strategy based web service selection using a learning-to-rank algorithm | |
| US20230342797A1 (en) | Object processing method based on time and value factors | |
| CN112418978A (en) | Product recommendation method, device, equipment and medium | |
| CN113761193A (en) | Log classification method and device, computer equipment and storage medium | |
| CN111738754A (en) | Object recommendation method and device, storage medium and computer equipment | |
| CN111651666A (en) | User theme recommendation method and device, computer equipment and storage medium | |
| CN114201626B (en) | Multimedia recommendation method and device, electronic equipment and storage medium | |
| US11977565B2 (en) | Automated data set enrichment, analysis, and visualization | |
| CN115630221A (en) | Terminal application interface display data processing method and device and computer equipment | |
| CN112925723B (en) | Test service recommendation method and device, computer equipment and storage medium | |
| CN113761134A (en) | User portrait construction method and device, computer equipment and storage medium | |
| WO2021129368A1 (en) | Method and apparatus for determining client type | |
| CN110929163B (en) | Course recommendation method and device, computer equipment and storage medium | |
| CN113850416A (en) | Advertisement promotion cooperation object determining method and device | |
| CN114238044A (en) | Method and device for calculating activity of open source project and computer equipment | |
| CN113724044A (en) | User portrait based commodity recommendation, apparatus, computer device and storage medium | |
| Kuznietsova et al. | Business Intelligence Techniques for Missing Data Imputations |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |